While a number of recent efforts are being made to achieve a finer pattern rule in the drive for higher integration and operating speeds in LSI devices, the commonly used light exposure lithography is approaching the essential limit of resolution determined by the light source wavelength.
As the light source used in the lithography for resist pattern formation, g-line (436 nm) or i-line (365 nm) from a mercury lamp has been widely used. One means believed effective for further reducing the feature size is to reduce the wavelength of exposure light. For the mass production process of 64 M-bit DRAM, the exposure light source of i-line (365 nm) was replaced by a KrF excimer laser having a shorter wavelength of 248 nm. However, for the fabrication of DRAM with a degree of integration of 1 G or more requiring a finer patterning technology (processing feature size 0.13 μm or less), a shorter wavelength light source is required. In particular, photolithography using ArF excimer laser light (193 nm) is now under investigation.
On the other hand, it is known in the art that the bilayer resist process is advantageous in forming a high-aspect ratio pattern on a stepped substrate. In order that a bilayer resist film be developable with a common alkaline developer, high molecular weight silicone compounds having hydrophilic groups such as hydroxyl and carboxyl groups must be used.
Among silicone base chemically amplified positive resist compositions, recently proposed were those compositions for KrF excimer laser exposure comprising a base resin in the form of polyhydroxybenzylsilsesquioxane, which is a stable alkali-soluble silicone polymer, in which some phenolic hydroxyl groups are protected with t-BOC groups, in combination with an acid generator (see JP-A 6-118651 and SPIE vol. 1925 (1993), p. 377). For ArF excimer laser exposure, positive resist compositions comprising as a base a silsesquioxane of the type in which cyclohexylcarboxylic acid has substituted thereon an acid labile group were proposed (see JP-A 10-324748, JP-A 11-302382, and SPIE vol. 3333 (1998), p. 62). For F2 laser exposure, positive resist compositions based on a silsesquioxane having hexafluoroisopropanol as a dissolvable group were proposed (see JP-A 2002-55456). The above polymer bears in its backbone a polysilsesquioxane containing a ladder skeleton produced through polycondensation of a trialkoxysilane or trihalosilane.
Silicon-containing (meth)acrylate polymers were proposed as a resist base polymer having silicon pendants on side chains (see JP-A 9-110938, J. Photopolymer Sci. and Technol., Vol. 9, No. 3 (1996), pp. 435-446).
The undercoat layer of the bilayer resist process is formed of a hydrocarbon compound which can be etched with oxygen gas, and must have high etching resistance since it serves as a mask when the underlying substrate is subsequently etched. For oxygen gas etching, the undercoat layer must be formed solely of a silicon atom-free hydrocarbon. To improve the line-width controllability of the upper layer of silicon-containing resist and to minimize the sidewall corrugation and pattern collapse by standing waves, the undercoat layer must also have the function of an antireflective coating. Specifically, the reflectance from the undercoat layer back into the resist film must be reduced to below 1%.
Now, the results of calculation of reflectance at film thickness varying up to the maximum of 500 nm are shown in FIGS. 1 and 2. Assume that the exposure wavelength is 193 nm, and the topcoat resist has an n value of 1.74 and a k value of 0.02. FIG. 1 shows substrate reflectance when the undercoat layer has a fixed k value of 0.3, the n value varies from 1.0 to 2.0 on the ordinate and the film thickness varies from 0 to 500 nm on the abscissa. Assuming that the undercoat layer of the bilayer resist process has a thickness of 300 nm or greater, optimum values at which the reflectance is reduced to or below 1% exist in the refractive index range of 1.6 to 1.9 which is approximate to or slightly higher than that of the topcoat resist.
FIG. 2 shows substrate reflectance when the undercoat layer has a fixed n value of 1.5 and the k value varies from 0.1 to 0.8. In the k value range of 0.24 to 0.15, the reflectance can be reduced to or below 1%. By contrast, the antireflective coating used in the form of a thin film of about 40 nm thick in the single-layer resist process has an optimum k value in the range of 0.4 to 0.5, which differs from the optimum k value of the undercoat layer used with a thickness of 300 nm or greater in the bilayer resist process. For the undercoat layer in the bilayer resist process, an undercoat layer having a lower k value, that is, more transparent is necessary.
As the material for forming an undercoat layer in 193 nm lithography, copolymers of polyhydroxystyrene with acrylates are under study as described in SPIE vol. 4345, p. 50 (2001). Polyhydroxystyrene has a very strong absorption at 193 nm and its k value is as high as around 0.6 by itself. By copolymerizing with an acrylic ester having a k value of almost 0, the k value of the copolymer is adjusted to around 0.25.
However, the resistance of the acrylic polymer to substrate etching is weak as compared with polyhydroxystyrene, and a considerable proportion of the acrylic ester must be copolymerized in order to reduce the k value. As a result, the resistance to substrate etching is considerably reduced. The etching resistance is not only reflected by the etching speed, but also evidenced by the development of surface roughness after etching. Through copolymerization of acrylic ester, the surface roughness after etching is increased to a level of serious concern.
Naphthalene ring is one of rings that have a more transparency at 193 nm and a higher etching resistance than benzene ring. JP-A 2002-14474 proposes an undercoat layer having a naphthalene or anthracene ring. However, since naphthol-copolycondensed novolac resin and polyvinyl naphthalene resin have k values in the range of 0.3 to 0.4, the target transparency corresponding to a k value of 0.1 to 0.3 is not reached, with a further improvement in transparency being necessary. The naphthol-copolycondensed novolac resin and polyvinyl naphthalene resin have low n values at 193 nm, as evidenced by a value of 1.4 for the naphthol-copolycondensed novolac resin and a value of only 1.2 for the polyvinyl naphthalene resin when the inventors measured. JP-A 2001-40293 and JP-A 2002-214777 describe acenaphthylene polymers which have lower n values and higher k values at the wavelength of 193 nm than at 248 nm, both falling outside the target values. There is a need for an undercoat layer having a high n value, a low k value, transparency and high etching resistance.
Also proposed was a tri-layer process of stacking a silicon-free single-layer resist as a topcoat, an intermediate layer containing silicon below the resist, and an organic undercoat below the intermediate layer. See J. Vac. Sci. Technol., 16(6), November/December 1979. Since the single-layer resist generally provides better resolution than the silicon-bearing resist, the tri-layer process permits such a high resolution single-layer resist to be used as an imaging layer for light exposure. A spin-on-glass (SOG) coating is used as the intermediate layer. A number of SOG coatings have been proposed.
In the tri-layer process, the optimum optical constants of the undercoat layer for controlling reflection from the substrate are different from those in the bilayer process. The purpose of minimizing substrate reflection, specifically to a level of 1% or less is the same between the bi- and tri-layer processes. In the bilayer process, only the undercoat layer is endowed with the antireflective effect. In the tri-layer process, either one or both of the intermediate layer and the undercoat layer may be endowed with the antireflective effect.
U.S. Pat. No. 6,506,497 and U.S. Pat. No. 6,420,088 disclose silicon-containing layer materials endowed with antireflective effect. In general, a multi-layer antireflective coating has greater antireflective effect than a single-layer antireflective coating and commercially widely used as an antireflective film for optical articles. A higher antireflective effect is obtainable by imparting an antireflective effect to both an intermediate layer and an undercoat layer. If the silicon-containing intermediate layer in the tri-layer process is endowed with the function of an antireflective coating, the undercoat layer need not necessarily possess the maximum function of an antireflective coating. In the tri-layer process, the undercoat layer is required to have high etching resistance during substrate processing rather than the antireflective coating function. Then a novolac resin having high etching resistance and containing more aromatic groups must be used as the undercoat layer in the tri-layer process.
FIG. 3 illustrates substrate reflectance with a change of the k value of the intermediate layer. It is seen that by setting a k value as low as 0.2 or less and an appropriate thickness to the intermediate layer, a satisfactory antireflective effect as demonstrated by a substrate reflectance of up to 1% is achievable. In general, the antireflective coating must have a k value of 0.2 or greater in order to reduce reflectance to or below 1% at a coating thickness of 100 nm or less (see FIG. 2). In the tri-layer structure wherein the undercoat layer serves to restrain reflection to a certain extent, the intermediate layer may have an optimum k value of less than 0.2.
FIGS. 4 and 5 illustrate changes of reflectance with the varying thickness of the intermediate layer and undercoat layer, when the undercoat layer has a k value of 0.2 and 0.6, respectively. The undercoat layer having a k value of 0.2 assumedly corresponds to the undercoat layer optimized for the bilayer process, and the k value of 0.6 for the undercoat layer is approximate to the k values at 193 nm of novolac and polyhydroxystyrene. The thickness of the undercoat layer varies with the topography of the substrate whereas the thickness of the intermediate layer is kept substantially unchanged so that presumably it can be coated to the predetermined thickness. The undercoat layer with a higher k value (0.6) is effective in reducing reflectance to 1% or less with a thinner film. In the event that the undercoat layer has a k value of 0.2 and a thickness of 250 nm, the intermediate layer must be increased in thickness in order to provide a reflectance of 1% or less. Increasing the thickness of the intermediate layer is not preferable because a greater load is applied to the topcoat resist layer during dry etching of the intermediate layer. To use the undercoat layer in thin film form, it must have not only a high k value, but also greater etching resistance.
In the event the underlying processable film is a low dielectric constant insulating film of porous silica, the positive resist suffers from a footing problem and a poisoning problem that scum is left in the spaces. It is believed that these problems arise because ammonia is adsorbed by the porous silica during the substrate cleaning step using ammonia and then liberated during the resist processing step to neutralize the acid generated in the exposed area of resist. JP-A 2004-177666 proposes to use a polymer having acidic groups for prohibiting the poisoning problem.